八皇后算法解析

今天研究力扣的一道题死活写不出来对应的算法,没办法自己算法基础太差。于是看了下答案,发现使用什么回溯算法,菜鸟表示平时开发期间写的最复杂的程序就是写了两层for循环,已经很牛逼了有木有?这个回溯算法什么鬼?于是乎百度了下,算是了解了回溯算法是什么玩意儿。这里分析一波八皇后算法来加深一下理解。

八皇后算法描述如下:在8×8格的国际象棋上摆放八个皇后,使其不能互相攻击,即任意两个皇后都不能处于同一行、同一列或同一斜线上,问有多少种摆法!

下面来分析一波,假设此时我们想要在黑色方块位置放置一个皇后:

如果一列一列的放置皇后的话,图中黑色位置能放置一个皇后的合法性条件为:

1、绿色线条经过的方格没有皇后 (不处于同一斜线)

2、红色线条经过的方格没有皇后 (不处于同一行)

3、紫色线条经过的方格没有皇后 (不处于同一斜线)
也就是说如果以黑色方块位置为参照原点:(0,0)坐标点,紫色和绿色两个线条分别是斜率为1和-1的两个函数,如下图:

紫色线所代表的函数是:y = -x;

绿色先所代表的函数是:y=x;

(横坐标是列,纵坐标为行,注意行从上到下递增)

凡是位于这两条函数线上的位置(点)以及横坐标(说明位于同一行)都不能有皇后。
所以假设某一列皇后的位置用行来记录,比如queen[column] = row,意思是第column列的皇后的位置在第row行。

同行的逻辑很好判断,那么我们想要在黑色方块位置放置一个皇后,怎么判断前面几列是否在绿色线条和紫色线条上已经有了皇后呢?思路也很简单:

假设黑色方块的位置为n列,nRow行,假设位于m列的所在的行是否有皇后位于紫色线或者绿色上,那么就符合下面条件:

1
2
3
4
5
6
7
8
9
10
11
//假设此时即将在n列放置一个皇后,n>m

]//获取m列上皇后所在的行
int mRow = queen[m]
int nRow = queen[n];
//行的差值
int rowDiff = nRow - mRow;

//列的差值
int columnDiff = n-m;

上面代码中 rowDiff的绝对值等于columnDiff的绝对值的话,说明点位于y=x或者y=-x的函数线上:

就说明此时黑色方块的位置是不能放置皇后的,因为在紫色或者绿色线上已经有了皇后。
那么用代码来(currentColumn,curreentRow)是否可以放置皇后的方法如下

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
     /**
* 判断当(currentRow,currentColumn)是否可以放置皇后
* @param currentColumn
* @return
*/
public boolean isLegal(int currentRow,int currentColumn) {
//遍历前面几列
for(int preColumn=0;preColumn<currentColumn;preColumn++) {
int row = queen[preColumn];
//说明在子preColumn的低currentRow已经有了皇后
if(row==currentRow) {
return false;
}

//行与行的差值
int rowDiff= Math.abs(row -currentRow);

//列于列的差值
int columnDiff = Math.abs(currentColumn-preColumn);
//说明斜线上有皇后
if(rowDiff==columnDiff ){
return false;
}
}//end for

//说明(currentRow,currentColumn)可以摆放。
return true;
}

}

因为博主是按照一列一列的方式来进行放置的,所以整体思路就是:在当前列逐步尝试每一行是否可以放置皇后,如果有一个可以放置皇后,就继续查看下一列的每一行是否可以放置皇后。所以代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
int queen[] = new int[8];
int count = 0;

private void eightQueen(int currentColumn) {
//这个for循环的目的是尝试讲皇后放在当前列的每一行
for(int row=0;row<8;row++) {
//判断当前列的row行是否能放置皇后
if(isLegal(row,currentColumn)) {
//放置皇后
queen[currentColumn] = row;
if(currentColumn!=7) {
//摆放下一列的皇后
eightQueen(currentColumn+1);

}else {
//递归结束,此时row要++了
count++;
}
}
}//end for
}

需要注意的是当currentColumn==7的时候,说明此时已经完成了一种摆放方法,然后for循环继续执行,去尝试其他摆放方法。

测试一波,一共有92种摆放方法:

1
2
3
4
5
6
public static void main(String args[]) {
Queen queen = new Queen();
queen.eightQueen(0);
System.out.println("总共有 " +queen.count+ " 摆放方法");
}

所以结合八皇后的实现来看看到底什么是回溯算法,看百度百科解释 (rel=undefined):回溯算法实际上一个类似枚举的搜索尝试过程,主要是<font color#ff00ff>在搜索尝试过程中寻找问题的解,当发现已不满足求解条件时,就“回溯”返回,尝试别的路径。但当探索到某一步时,发现原先选择并不优或达不到目标,就退回一步重新选择,这种走不通就退回再走的技术为回溯法

比如八皇后算法来说,我们根据约束条件判断某一列的某一行是否可以放置皇后,如果不可以就继续判断<font color #ff00ff>当前列的下一行是否可以放置皇后.如果可以放置皇后,就继续探寻下一列中可以放置皇后的那个位置。完成一次摆放后。再重新挨个尝试下一个可能的摆放方法。

下面用一个力扣的题 (rel=undefined)再次巩固下回溯算法的应用。该题描述如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
给定一个无重复元素的数组 candidates 和一个目标数 target ,找出 candidates 中所有可以使数字和为 target 的组合。candidates 中的数字可以无限制重复被选取。

说明:所有数字(包括 target)都是正整数。解集不能包含重复的组合。
示例 1:输入: candidates = [2,3,6,7], target = 7,
所求解集为:
[
[7],
[2,2,3]
]

示例 2:输入: candidates = [2,3,5], target = 8,
所求解集为:
[
[2,2,2,2],
[2,3,3],
[3,5]
]

做该题的重要条件是无重复的数组,那么问题就很好解了。

首先对数组从大到小排序。这是解题的关键。

然后为了减少不必要的遍历,我们要对原来的数组进行截取:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
List<List<Integer>> res = new ArrayList<>();
//重要的要大小排列
Arrays.sort(candidates);
//说明原数组中就没有满足target的数
if (candidates[0] > target) {
return res;
}

List<Integer> newCandidates= new ArrayList<Integer>();
int len = candidates.length;
// 取小于target的数 组成一个临时数组
for (int i = 0; i < len; i++) {
int num = candidates[i];
if (num > target) {
break;
}
newCandidates.add(num);
} // end for

通过上面的步骤我们拿到了一个从小到大排列的无重复数组newCandidates,数组中的元素都<=target.

因为数组从小到大排列,所以我们有如下几种情况,以candidates = [2,3,5], target = 8为例:

符合条件的子数组满足条件如下

1、target循环减去一个数,如果能一直减到到差值等于0,那么这个数组成的数组就是一个解,比如[2,2,2,2];

2、target减去一个数,然后形成了一个新的newTarget=target-num[i],让这个newTarget减去下一个数num[i+1],然后执行步骤1,则又是一个解,比如[2,3,3];(其实步骤1是步骤2的一个特例)

3、target减去一个数,然后形成了一个新的newTarget=target-num[i],让这个newTarget减去下一个数num[i+1],如果能一直减到到差值等于0说明又是一个解.,比如[3,5];

如此得到了一个规律,只要是相减之后得到差值=0,就说明就得到一个解。

得到一个新的解之后继续循环数组中的下一个数字,继续执行1,2,3步骤即可。

所以完成的解法如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
class Solution {
public List<List<Integer>> combinationSum(int[] candidates, int target) {
List<List<Integer>> res = new ArrayList<>();
//重要的要大小排列
Arrays.sort(candidates);

List<Integer> temp = new ArrayList<Integer>();


if (candidates[0] > target) {
return res;
}

int len = candidates.length;

// 取小于target的数 足证一个临时数组
for (int i = 0; i < len; i++) {
int num = candidates[i];
if (num > target) {
break;
}

temp.add(num);
} // end for

//
find(res, new ArrayList<>(), temp, target, 0);

return res;
}

public void find(List<List<Integer>> res, List<Integer> tmp, List<Integer> candidates, int target, int start){
//target==0.找到一个新的解
if (target == 0) {
res.add(new ArrayList<>(tmp));
}else if(target>0){
for (int i = start; i < candidates.size(); i++) {
int num = candidates.get(i);
if(num<=target){
tmp.add(num);
//查找新的target
int newTarget = target-num;
find(res, tmp, candidates, newTarget, i);
tmp.remove(tmp.size() - 1);
}

}//end for
}
}
}

原文:https://blog.csdn.net/chunqiuwei/article/details/90113087#commentBox